We propose to continue to conduct a case-control study to test the hypothesis that persons with relatively high levels of selenium are at decreased risk of developing cancer of the breast, lung, and large bowel. We have recently expanded the study to include cancer of the prostate, bladder, and ovary. The study enrolls nearly all incident cases that occur in the nine-county region surrounding Rapid City, SD, and a random sample of general population controls from this area. Selenium intake is measured by neutron activation analysis of toenail clippings from all subjects, providing an assessment of selenium status approximately one year previously. Information on potential modifiers and confounders, including other dietary variables, is collected by interview and questionnaire. Results from other studies using prospectively collected specimens suggest that low levels of selenium may confer an increased risk of cancer in humans. However, it is not known whether only individuals at the lowest extreme are at increased risk, or whether high levels (and therefore perhaps supplementation) confers protection. The Rapid City region provides an ideal setting because the selenium levels there are among the highest in the country, and the variation among individuals is unusually large. The current study has the advantages of an existing smooth and efficient operation, a high degree of cooperation from participants and health care providers, and a population based design with nearly complete ascertainment of cases. To date over 280 cases and over 160 controls have been enrolled, with cooperation rates of 98% and 93%, respectively. Initial results show a strong protective trend for lung cancer, but no significant association with cancer of the breast and large bowel. However, the 95% confidence limits will still include relative risks of potential importance which cannot be excluded without expanding the present study. Since tissue selenium levels are easily increased by supplementation, a precise estimate of associations with cancer risk have considerable potential practical relevance. We thus propose to double the sample size of this study to better quantify these relatioships.